Hi HN!,While discussing agentic engineering with my close network, it came to light that most agent skills still live on desktops, in repositories, or in people’s heads.In the context of large, regulated organisations, where compliance and governance are key drivers, it makes sense for skills to be treated like any other governed, trusted and compliant data asset.From a software and agentic engineering perspective, the tooling should be easy to adopt and make agentic coding easier, faster, safer .This started to shape a new capability, “Skills as a Service”, delivered by the Coding Agent Skill Library, which moves skills into a secure, governed and trusted service.It has two flavours:1. Drop-in library: If you don t want to change your current workflow, pull any skills you need directly into your environment and use them with Claude, Codex, Cursor, Antigravity, and other agent frameworks or IDEs.2. Service via MCP: If you want Skills as a Service, then Agents and IDEs access a semantically linked, ontology-first knowledge graph through a read-only MCP server. This is where the real value is.The knowledge graph is designed to retrieve the right skills for the shape of the task, rather than relying only on names, folders or basic metadata. It models task intent, workflow stage, supporting evidence, skill versions and relationships between skills.Runs locally in docker, wired into VSCode or Cursor using stdio. Intent is this would be scaled out for enterprise environments, have inner source style contributions federated ops model. My thinking was to enable controlled contribution, consistent quality, compliance and simplify skill discovery and reduce drift, by having a single trusted source.The aim was simple: make curated skills a safe, versioned and governed organisational asset, accessible from a single internal service. Question is does it actually help make real builders d2d - faster, safer, cheaper, easier and does it help enterprises govern skills usage?Repo: https://github.com/mmccalla/coding-agent-skill-library